Reliable modeling of adobe masonry requires material-specific characterization of the weak and highly variable adobe–mortar interface; however, experimentally calibrated interface failure envelopes for earthen masonry remain limited. This study presents a practical and systematic experimental–calibration procedure for characterizing adobe–mortar interaction under multiple stress states and translating the measured response into parameters suitable for numerical modeling. Four mortar configurations were investigated: plain mud mortar, mud mortar reinforced with tire-production waste fibers, and mud mortars reinforced with wheat straw fibers soaked for 10 min and one week. Direct tension, direct shear, compression, and combined compression–shear tests were conducted, and the results were represented by a unified multi-surface failure framework comprising a tensile cut-off, a frictional shear branch, and an elliptical compressive cap. Mortar composition and fiber condition affected both interface strength and the calibrated model parameters. The one-week-soaked straw mortar produced the highest mean measured strengths, with direct tensile, direct-shear, and compressive strengths of 0.040, 0.020, and 4.7 MPa, respectively, compared with 0.009, 0.008, and 3.5 MPa for plain mud mortar, while the fiber-containing mortars showed lower scatter than plain mud within the investigated three-specimen sets. The calibrated parameters are specific to the tested soil, materials, and loading protocol; nevertheless, the proposed procedure provides a transferable framework for the material-specific calibration of other earthen masonry systems. With validation across a broader range of soils, mortars, fiber conditions, and loading protocols, this approach could support comparative databases, design-oriented relationships, and the future development of technical guidance for adobe masonry interfaces.